MODELING NON-LINEAR RELATIONSHIPS IN ROCK MASS RATING (RMR) SYSTEM: A METHODOLOGICAL STACKING-BASED HYBRID APPROACH

Öz

Rock Mass Rating (RMR) is a fundamental classification system in rock engineering. However, the traditional RMR method relies on a simple arithmetic summation of parameters, often failing to fully capture complex non-linear interactions and field uncertainties. To overcome these limitations, this study proposes a novel stacking-based hybrid prediction model. Three machine learning algorithms with distinct paradigms—Random Forest (RF), Gradient Boosting (XGBoost), and Support Vector Regression (SVR)—were optimized via Bayesian Optimization. They were trained on a 2000-sample dataset generated through Monte Carlo simulation, which strictly preserved physical correlations. The proposed hybrid model achieved high predictive accuracy (R2 = 0.928, RMSE = 4.56), demonstrating significant superiority over the traditional RMR method (R2 = 0.842, RMSE = 6.76). Furthermore, feature importance analysis challenged conventional assumptions by identifying "discontinuity condition" as the most critical parameter affecting the RMR score, followed by rock quality designation (RQD). Ultimately, this study strongly supports that multi-model hybrid architectures offer a more flexible, precise, and reliable alternative to classical summation methods in rock mechanics.

Anahtar Kelimeler

• Rock Mass Rating (RMR)
• Machine Learning
• Stacking.
• Monte Carlo simulation
• Bayesian optimization

KAYA KÜTLE DERECESİ (RMR) SİSTEMİNDE DOĞRUSAL OLMAYAN İLİŞKİLERİN MODELLENMESİ: METODOLOJİK YIĞINLAMA TABANLI HİBRİT YAKLAŞIM

Öz

Kaya Kütle Derecesi (RMR), kaya mühendisliğinde en yaygın kullanılan sınıflandırma sistemlerinden biridir. Ancak geleneksel RMR yöntemi, parametrelerin basit aritmetik toplamına dayandığı için doğadaki karmaşık ve doğrusal olmayan ilişkileri tam olarak yansıtamamaktadır. Bu çalışmada, klasik yöntemlerin sınırlarını aşmak amacıyla yığınlama (stacking) tabanlı hibrit bir tahmin modeli geliştirilmiştir. Çalışma kapsamında; hiperparametreleri Bayesian optimizasyon ile ayarlanan Rastgele Ormanlar (RF), Gradyan Artırma (XGBoost) ve Destek Vektör Regresyonu (SVR) algoritmaları kullanılmıştır. Bu modeller, fiziksel etkileşimlerin korunduğu ve saha belirsizliklerinin Monte Carlo simülasyonu ile modellendiği 2000 adetlik sentetik bir veri seti üzerinde eğitilmiştir. Elde edilen bulgulara göre, geliştirilen hibrit model (R2 = 0,928 ve RMSE = 4,56), geleneksel RMR yöntemine (R2 = 0,842 ve RMSE = 6,76) kıyasla belirgin bir üstünlük sağlamıştır. Ayrıca öznitelik önem analizinde, literatürdeki genel kanının aksine RMR puanını en çok etkileyen parametrenin "süreksizlik durumu" olduğu, bunu "kaya kalite indeksi (RQD)"nin izlediği tespit edilmiştir. Sonuç olarak bu çalışma; kaya mekaniğinde çoklu model mimarisine dayalı hibrit yaklaşımların, klasik toplama yöntemlerine göre daha esnek, hassas ve güvenilir bir alternatif sunduğunu güçlü biçimde desteklemektedir.

Anahtar Kelimeler

• Kaya Kütle Derecesi (RMR)
• Makine Öğrenimi
• Yığınlama
• Monte Carlo Simulasyonu
• Bayesian Optimizasyonu

Kaynakça

1. Kaynaklar (References)
2. Barton, N., Lien R., Lunde, J., 1974. Engineering classification of rock masses for the design of tunnel support. Rock Mechanics 6, 189–236. https://doi.org/10.1007/BF01239496.
3. Bieniawski, Z. T., 1973. Engineering classification of jointed rock masses. Transactions of the South African Institution of Civil Engineers, 15(12), 335-344.
4. Bieniawski, Z. T., 1989. Engineering rock mass classifications: A complete manual for engineers and geologists in mining, civil, and petroleum engineering. John Wiley & Sons.
5. Breiman, L., 2001. Random forests. Machine Learning, 45(1), 5-32. https://doi.org/10.1023/A:1010933404324.
6. Chen, T., Guestrin C., 2016. XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ’16, Association for Computing Machinery, New York, NY, USA, pp. 785-794, https://doi.org/10.1145/2939672.2939785.
7. Cemiloglu, A., Zhu, L., Arslan, S., Xu, J., Yuan, X., Azarafza, M., Derakhshani, R., 2023. Support Vector Machine (SVM) Application for Uniaxial Compression Strength (UCS) Prediction: A Case Study for Maragheh Limestone. Applied Sciences, 13(4), 2217. https://doi.org/10.3390/app13042217.
8. Hansen, T. F., Aarset, A., 2024. Unsupervised machine learning for data-driven rock mass classification: Addressing limitations in existing systems using drilling data. Rock Mechanics and Rock Engineering, 57(10), 7705-7735. https://doi.org/10.1007/s00603-024-04280-z.

9. Hansen, T. F., Erharter, G. H., Liu, Z., Torresen, J., 2024. A comparative study on machine learning approaches for rock mass classification using drilling data. Applied Computing and Geosciences, 24, 100199. https://doi.org/10.1016/j.acags.2024.100199.
10. Hu, J., Zhou, T., Ma, S., Yang, D., Guo, M., Huang, P., 2022. Rock mass classification prediction model using heuristic algorithms and support vector machines: a case study of Chambishi copper mine. Scientific Reports, 12(1), 928. https://doi.org/10.1038/s41598-022-05027-y.
11. Jia, Z. C., Wang, Y., Wang, J. H., Pei, Q. Y., Zhang, Y. Q., 2024. Rockburst intensity grade prediction based on data preprocessing techniques and multi-model ensemble learning algorithms. Rock Mechanics and Rock Engineering, 57(7), 5207-5227. https://doi.org/10.1007/s00603-024-03811-y.
12. K, F.A.S., Tripathy, D.P., Nanda, S.K., 2025. Machine learning-based predictive models for enhancing safety and preventing failures in mining operations. Safety in Extreme Environments, 7(20). https://doi.org/10.1007/s42797-025-00130-5.
13. Katuwal, T.B., Panthi, K.K., Basnet, C.B., 2025. Machine Learning Approach for Rock Mass Classification with Imbalanced Database of TBM Tunnelling in Himalayan Geology. Rock Mechanics and Rock Engineering, 58, 11293–11318. https://doi.org/10.1007/s00603-024-04212-x.
14. Khajehzadeh, M., Keawsawasvong, S., Motahari, M. R., Jamsawang, P., 2024. Effective machine-learning models for rock mass deformation modulus estimation based on rock mass classification systems. Engineered Science, 29, 1120. https://doi.org/10.30919/es1120.
15. Maier, H. R., Galelli, S., Razavi, S., Castelletti, A., Rizzoli, A., Athanasiadis, I. N., Miquel Sànchez-Marrè, M., Acutis, M., Wu, W., Humphrey, G. B., 2023. Exploding the myths: An introduction to artificial neural networks for prediction and forecasting. Environmental modelling & software, 167, 105776. https://doi.org/10.1016/j.envsoft.2023.105776.
16. Nabavi, Z., Hosseini, S., Hamidi, J. K., Monjezi, M., Hasheminasab, F., Vardin, A. N., Hasanipanah, M., 2025. Prediction of Rotary Drilling Rate of Penetration in Surface Mines Using Machine Learning Techniques. Rock Mechanics and Rock Engineering, 1-31. https://doi.org/10.1007/s00603-025-05014-5.
17. Rao, G., Rao, Y., Wan, J., Huang, Q., Xie, Y., Lai, Q., Yang, Z., Xiang, R., Zhang, L., 2024. Long-term prediction modeling of shallow rockburst with small dataset based on machine learning. Scientific Reports, 14(1), 16131. https://doi.org/10.1038/s41598-024-64107-3.
18. Shahani, N. M., Kamran, M., Zheng, X., Liu, C., Guo, X., 2021. Application of gradient boosting machine learning algorithms to predict uniaxial compressive strength of soft sedimentary rocks at Thar Coalfield. Advances in Civil Engineering, 2021(1), 2565488. https://doi.org/10.1155/2021/2565488.
19. Singh, J., Pradhan, S.P., Singh, M.P., 2025. Application of machine learning tools in predicting basic rock mass rating of mountainous road cut slopes. Journal of Earth System Science, 134(52). https://doi.org/10.1007/s12040-024-02515-9.
20. Vapnik, V., 2013. The nature of statistical learning theory. Springer science & business media. https://doi.org/10.1007/978-1-4757-3264-1.
21. Yang, B., Jahed Armaghani, D., Fattahi, H., Afrazi, M., Koopialipoor, M., Asteris, P. G., Khandelwal, M., 2025. Optimized random forest models for rock mass classification in tunnel construction. Geosciences, 15(2), 47. https://doi.org/10.3390/geosciences15020047.
22. Yang, W., Zhao, J., Li, J., Chen Z., 2023. Probabilistic machine learning approach to predict incompetent rock masses in TBM construction. Acta Geotechnica, 18, 4973–4991. https://doi.org/10.1007/s11440-023-01871-y.
23. Yu, B., Wang, B., & Zhang, Y., 2024. Application of artificial intelligence in coal mine ultra-deep roadway engineering—a review. Artificial Intelligence Review, 57(262). https://doi.org/10.1007/s10462-024-10898-w.
24. Zhang, Q., Li, S., Ling, J., Liu, Z., 2025. Prediction of TBM Performance Under Imbalanced Geological Conditions Using Resampling and Data Synthesis Techniques. Applied Sciences, 15(22), 12065. https://doi.org/10.3390/app152212065.
25. Zvarivadza, T., Lawal, A. I., Onifade, M., Mulenga, F., Kwon, S., Khandelwal, M., 2025. Deep Learning-Powered Rock Mass Classification: Predicting RMR from Q-System Parameters with High Accuracy. Rock Mechanics Bulletin, 100219. https://doi.org/10.1016/j.rockmb.2025.100219.